20 Korean Particles and Endings, Ranked by How Often They Actually Appear

There are 326 documented grammar patterns in the reference we build our grammar pages from. Nobody learns 326 of anything, and the usual advice — start with the basics — quietly assumes you already know which ones those are.

So we checked. Every pattern in that reference was counted against 39,678 lines of transcripts from the clips in our shadowing library, and the 41 below are the ones that actually turned up often enough to matter.

That is the useful number here: not how important a grammar book says a particle is, but how many lines of real speech it appeared in.

Each entry links to its full explanation with examples. This page is the map; those pages are the territory.

What we could count, and what we deliberately skipped

This section exists because a ranking is only worth as much as its method, and ours has a firm boundary we decided not to cross.

A line is split into words, punctuation is stripped, and a word counts as a match if it ends with the pattern. Polite and casual forms of the same pattern are added together, so -더라고 and -더라고요 land in one figure rather than two.

Patterns written with bare jamo were excluded entirely. A form like -(으)ㄹ까요 attaches its ㄹ as a final consonant on the previous syllable — 갈까요, not ㄹ까요 — so no word ever literally ends with the written pattern. Counting it this way returns zero, which is not a small error but a completely wrong answer. Multi-word patterns such as -고 있다 were excluded for a related reason: matching them requires searching whole lines, which overcounts badly.

Before trusting any of this we ran the counter against 28 patterns that had already been counted and hand-checked by a separate script. Eighteen of the twenty countable ones matched within ten percent, and most matched exactly — 는데 1,211 against 1,207, 니까 978 against 978, 더라고요 70 against 70.

The two that did not match are worth naming. -아/어야지 and -아/어야겠다 came in slightly low, because forms built on 하다 contract to 해야지 rather than keeping the 아 or 어 our patterns look for. Both are undercounts of a few lines, and both are in the lower half of this list.

One precision limit remains and it runs the other way. A particle that is spelled like an ordinary word gets credit for both. 같이 is counted here as a particle meaning like, but the same spelling is an everyday adverb meaning together, and this method cannot tell them apart. Read its figure as generous.

You will meet these constantly — 200 lines and up

Eight patterns, and between them they account for more lines than everything else on this page combined.

Most of them are particles rather than endings, which is worth noticing. They attach to nouns and declare what role that noun is playing — where something happened, who it went to, what it started from, what it is being compared against.

Subtitles handle these by rewriting the sentence into English word order, so the particle does its job and then disappears from the translation without ever being taught to you.

If you learn nothing else here, learn these eight. They are short, they are regular, and they are load-bearing.

• 잖아 (jana) — Indicates that the speaker is reminding the listener of a reason they should be aware or might have forgotten. — counted in 828 lines. Full explanation.
• 하고 (hago) — Connects multiple nouns, indicating a list or group without specifying order. — counted in 539 lines. Full explanation.
• 에서 (eseo) — Used to indicate where an action occurs or to indicate a reference — counted in 427 lines. Full explanation.
• 거든(요) (geodeun) — Referring to the reason, cause, or basis for the preceding statement that the speaker thought of. — counted in 279 lines. Full explanation.
• 부터 (buteo) — Indicate the starting point of an action, time or place. — counted in 272 lines. Full explanation.
• 한테 (hante) — Define a person or a living creature as the recipient or target of an action. — counted in 258 lines. Full explanation.
• 같이 (gachi) — Indicates resemblance or a metaphorical comparison in actions or objects — counted in 248 lines. Full explanation.
• 까지 (kkaji) — Indicate the end of the range of time, place or action. — counted in 242 lines. Full explanation.

Frequent enough to learn deliberately — 100 to 200 lines

This tier is where a sentence stops being a list of words and starts having a shape.

Several of these connect clauses rather than mark nouns — giving a reason, drawing a contrast, setting one action against another. They are the joins, and a learner who knows only the nouns and verbs hears a sequence of fragments where a native listener hears one argument.

They are also more forgiving than the first tier. Missing one usually costs you nuance rather than meaning, so recognition here is genuinely enough for a while.

A practical approach: learn to spot them, and worry about producing them only once you catch all of them reliably.

• 아/어서 (eoseo) — The first clause is the reason for or the cause of the second clause. — counted in 187 lines. Full explanation.
• 보고 (bogo) — Used to indicate the person to whom a question, request, order, etc is directed at. Commonly used in colloquial speech when quoting others. — counted in 152 lines. Full explanation.
• 보다 (boda) — Express that something is more or less than the noun in a certain quality or quantity. — counted in 147 lines. Full explanation.
• 이다 (ida) — Express that the subject and predicate are the same thing. — counted in 147 lines. Full explanation.
• 대로 (daero) — When added to the end of a noun, it indicates that something is "as" or "in accordance with" the noun. — counted in 146 lines. Full explanation.
• 하다 (hada) — Turns nouns into action verbs, turns nouns into descriptive verbs, to perform actions, to do something. — counted in 134 lines. Full explanation.
• 다가 (daga) — Expresses the speaker's interruption of one action to immediately start another action. — counted in 132 lines. Full explanation.
• 아/어야 (eoya) — Express a necessity, obligation or a condition that has be completed for something else to realize. — counted in 127 lines. Full explanation.
• 지만 (jiman) — Express a contrast between two clauses and the second clause is usually a contrast to the first clause. — counted in 124 lines. Full explanation.
• 랑/이랑 (irang) — Connects multiple nouns, indicating a list or group without specifying order. — counted in 123 lines. Full explanation.
• 때문에 (ttaemune) — Expresses a reason or cause of a situation. — counted in 104 lines. Full explanation.
• 기는(요) (gineun) — Expression is used to politely refute or disagree with the other person's statement. Used in response to a compliment to show modesty. Can also be used lightly criticize or scold someone. — counted in 101 lines. Full explanation.

The rest of this set

This page covers 20 of 41. The rest continue here.

• 20 Korean Particles and Endings, Ranked by How Often They Actually Appear (this page)
• Worth recognising — Korean grammar (21 entries)

Particles do the work English does with word order

If this list looks particle-heavy compared to a textbook's early chapters, that is not an accident of our corpus.

English marks who did what to whom mostly by position: the subject comes before the verb, the object after. Korean marks it with a tag attached to the noun, which frees word order and puts the grammatical load on a handful of short syllables.

That is why missing one particle can flip a sentence in a way that missing an adjective never does. The words are all still there and the relationship between them is gone.

It is also why particles reward early attention out of proportion to how small they look on the page.

A reasonable order to work through this

Start with the first section and do not skip ahead. Eight patterns is a realistic week, and every one of them will appear in the next thing you watch.

For each, open its explanation page and read one example aloud rather than reading the definition twice. The definition tells you what it does; the example tells you where it sits in a sentence, which is the part that transfers.

The second section is where the work stops being obvious and starts being worthwhile. These appear often enough to matter and rarely enough that you can go a while without noticing you never learned them.

Treat the third section as recognition only. Knowing that a pattern exists and roughly what it signals is enough to stop it from blocking a sentence, and that is the entire job at this stage.

If a pattern refuses to stick, check whether the problem is actually pronunciation — several of these change shape once attached, and we listed the words most likely to slip past your ear separately.

Source and licence

The grammar definitions and pattern inventory on this page come from kimchi-grammar, an open Korean grammar reference, used under CC BY 4.0. The definitions are theirs; we have not altered their wording.

The frequency figures are ours. They were measured against our own clip transcripts using the method described above, and they are not part of the kimchi-grammar data.

Our per-pattern explanation pages carry the same attribution.

FAQ

Why only 41 of 326? Two filters. A pattern had to appear in at least twenty lines of our transcripts, and it had to be one our method can count reliably — which excludes patterns written with bare jamo and multi-word constructions. The excluded ones are not unimportant; they are uncountable by this method.

Are these counts exact? Close but not exact, and the method section is specific about where they drift. Word-final matching is reliable, contracted 하다 forms are undercounted slightly, and particles spelled like ordinary words are overcounted.

Is this the same as your endings list? No, and they complement each other. The other page ranks sentence endings from a hand-checked candidate list; this one sweeps the whole grammar reference and comes out particle-heavy as a result.

Does idol content skew the ranking? For particles, barely — they are structural and appear in any Korean. For endings the register matters more, since casual speech favours different endings than writing does. Treat the particle ranking as general and the ending ranking as conversational.

Where do the explanations come from? From kimchi-grammar, an open reference used under CC BY 4.0, credited in full at the bottom of this page. The frequency measurements are ours.